WP1: Live issues (Lead: Bernd Stahl)
Aim: To understand the practical issues that organisations, particularly SMEs, face when developing products and services that integrate or are based on generative AI.
Method: This WP will draw on the ideas of responsible AI ecosystems by providing an initial delineation of relevant ecosystems. To understand which issues an organisation faces, it will be imperative to determine the boundaries of the system in question. This may be an organisation, an industry, a particular jurisdiction, or a project. The understanding of the boundaries of the ecosystem in question informs insights on the benefits that are expected and the concerns that may arise. Each generative AI ecosystem will be described using a case study approach. We expect to investigate a total of eight case study ecosystems (5 in the UK, 3 in Africa). These will be
identified through searches of new generative AI business models, as publicised by incubators, in trade conferences, industry publications. RAI-UK resources and network partners will be considered as possible sources of case studies. For each case, a minimum of 3 interviews will be undertaken
which will normally include an ecosystem leader, a technical expert and a potential user. Further information will be collected from websites, organisational reports etc
WP2: Co-creation and agile updates of SME guidance (Lead: Damian Eke)
Aim: To provide actionable guidance that will help SMEs implement responsible AI practices by ensuring adequate attention to relevant issues when developing/using generative AI-based products or services.
Method: the WP will employ an agile co-creation approach to develop relevant guidance sensitive to both UK and African socio-cultural and business contexts. The suitability of this draft guidance process and its fitness for purpose will be checked with relevant experts. This will form part of the interviews undertaken for WP1 which will allow for the interpretation of the guidance in the case context. By tying the co-creation process into the case research work, we will be able to make use of the update cycles, thus ensuring that our work remains current, and we keep abreast of technical
and regulatory developments. This guidance will build on existing frameworks for Responsible AI design and deployment (including but not limited to the OECD AI Principles, Trustworthy AI principles, EU AI Act, UK Algorithmic Transparency Recording Standard , Edinburgh declaration on responsibility for responsible AI). Overall, RAISE’s approach is a bottom-up approach where relevant stakeholders provide insights to contribute to RAISE’s guidance. The following methods will be used:
- Co-creation workshops (stakeholders will be drawn from the project use cases, SMEs in UK and Africa, academia, and civil society groups).
- Human-centric design workshops (HCD) (2x Virtual; 1 in person) – participants will engage in activities like persona creation, user-storyboarding and application of responsible innovation frameworks to create guidelines that meet core AI ethical and legal principles
- Wider community engagement workshop (1x virtual) – participants will include policy makers and civil society groups to review developed guidance and provide insights for adjustments
- Co-creation and guidance development and testing in Africa will primarily be undertaken using telecommunication technologies. However, to understand local requirements and engage with African SMEs, two field trips to Africa (Nigeria, Kenya, South Africa) are planned.
WP3: Testing, Implementation, evaluation, and future proofing (Lead: David Barnard-Wills)
Aim: To test the guidance processes developed in WP2, in practice with a software development team in a commercial SME context. This WP will Investigate how this guidance could be implemented with the Sociotech for Good (STG) team at Trilateral and use this experience to evaluate processes and to shape them in ways that render them stable and future proof.
WP4: Outreach and exploitation (Lead: Rowena Rodrigues)
Aim: Considering the world-wide reach and potential impact of generative AI, the impact we generate needs to be broader than the organisations engaged in the RAISE project. This WP will multiply impact by reaching out to potential user SMEs, industry groups and policymakers affected by and/or shape current and emerging generative AI ecosystems and practice.
Method: Outreach and exploitation will focus on boosting the project’s
- conceptual impact – contributing to the understanding of live issues and reframing debates on generative AI issues for SMEs
- instrumental impact – influencing the development of responsible AI policy, practice or services in SMEs
- knowledge and capacity building impact – practically helping SMEs to address generative AI issues